Loading ToolBench
MCP server for collaborative content structuring that transforms sparse user input into full, engaging articles
Blaster has well-structured tool definitions with complete input schemas and clear descriptions. All 5 tools follow verb_noun naming (generate_outline, expand_section, check_completeness, score_engagement, format_for_platform). Descriptions are substantive (100-180 chars), exceeding the 10-char minimum. Input schemas use proper JSON Schema with types, enums, and constraints. However, output schemas are not documented, LLMs cannot predict response structure for downstream chaining. Error handling is present but generic; no recovery guidance or actionable error messages visible in tool definitions. Tool annotations (readOnlyHint, destructiveHint) are absent despite all tools being read-only operations.
Check article content for completeness. Identifies thin sections needing expansion, missing elements from outline, and provides a completeness score.
Expand a brief section description into full content. Transforms a section title and description into well-written content with examples and smooth transitions.
Format content for a specific publishing platform. Optimizes for LinkedIn, Medium, Dev.to, Twitter, or generic use with appropriate formatting, hashtags, and character limits.
Generate a structured article outline from a topic. Transforms a topic and key points into a structured outline with sections, suggested content, and word count targets.
Score content for engagement quality. Analyzes hook effectiveness, flow, takeaways, headline, and structure. Returns detailed scores and improvement suggestions.
Output schemas not documented. LLMs cannot predict response structure, breaking tool chaining and forcing agents to guess field names for downstream calls.
Tool annotations missing. All 5 tools are read-only (no state modification), but readOnlyHint is not declared in tool definitions. Agents cannot distinguish safe tools from destructive ones.
Error handling lacks recovery guidance. Error responses in error_handler.py are generic; no actionable next steps or suggestions for LLM self-correction.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 58 | <=2025-11-25 | v2 |
Parameter 'outline' in check_completeness is optional but its nested schema (title, hook, sections, etc.) lacks descriptions. LLMs cannot understand what each field controls.
No pagination or result limits documented. Tools like generate_outline and expand_section may return large content blocks; no guidance on chunking or streaming.